> From: Robert Seeberger <[EMAIL PROTECTED]> > > From: > > > > Out of curiosity, why do you think you are in a position to mock > techniques > > that are fairly standard in science as well as economics? The > normalizing > > out of known uninteresting variations, like seasonal variations, > that do > > not help one answer a question one is interested is very standard. > > > > In this case, the change in the basic employment picture is of > greatest > > interest. Since seasonal variations exist, and are not indicative > of real > > trends, any trend analysis needs to normalize out this variation. > It would > > be similar to normalizing/subtracting out a known time dependant > background > > from a physical signal. > > > > If you are right, then some very successful scientific techniques > must be > > bogus. If they are bogus, then the obvious question is "why do they > work?" > > > > Maybe I am just misunderstanding. > > "The advance number of actual initial claims under state programs, > unadjusted, totaled 516,501 in the week ending Dec. 27, an increase of > 91,785 from the previous week. There were 620,929 initial claims in > the comparable week in 2002." > > It looks like a comparison between the previous week and then with the > same week in the previous year. Where would you get a seasonal > correction out of that? > > I would think that most people would read the sentence the same way > the Fool did. I certainly did/do. > > What is it I am missing here?
The statistical model assumes that hiring will increase before christmas. But that's not what's actually happening in reality as more people are losing their jobs for the past several weeks. Since seasonal patterns are being disrupted, the number becomes distorted. That's what happened this year. That what happened last year. That's what happens with bad statistical models. _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
